git-vuln-finder/git_vuln_finder/vulnerability.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
#
# Finding potential software vulnerabilities from git commit messages
#
# Software is free software released under the "GNU Affero General Public License v3.0"
#
# This software is part of cve-search.org
#
# Copyright (c) 2019-2020 Alexandre Dulaunoy - a@foo.be

import re
import sys
from langdetect import detect as langdetect


def find_vuln(commit, pattern, verbose=False):
    """Find a potential vulnerability from a commit message thanks to a regex
    pattern.
    """
    m = pattern.search(commit.message)
    if m:
        if verbose:
            print("Match found: {}".format(m.group(0)), file=sys.stderr)
            print(commit.message, file=sys.stderr)
            print("---", file=sys.stderr)
        ret = {}
        ret["commit"] = commit
        ret["match"] = m.groups()
        return ret
    else:
        return None

def find_vuln_event(commit_msg, pattern, verbose=False):
    """Find a potential vulnerability from a commit message thanks to a regex
    pattern.
    """
    m = pattern.search(commit_msg)
    if m:
        if verbose:
            print("Match found: {}".format(m.group(0)), file=sys.stderr)
            print(commit_msg, file=sys.stderr)
            print("---", file=sys.stderr)
        ret = {}
        ret["commit"] = commit_msg
        ret["match"] = m.groups()
        return ret
    else:
        return None


def summary(
    repo,
    commit,
    branch,
    tagmap,
    pattern,
    origin=None,
    vuln_match=None,
    tags_matching=False,
    commit_state="under-review",
):
    potential_vulnerabilities = {}
    rcommit = commit
    cve, cve_found = extract_cve(rcommit.message)
    if origin is not None:
        origin = origin
        if origin.find("github.com"):
            origin_github_api = origin.split(":")[1]
            (org_name, repo_name) = origin_github_api.split("/", 1)
            if repo_name.find(".git$"):
                repo_name = re.sub(r".git$", "", repo_name)
            origin_github_api = "https://api.github.com/repos/{}/{}/commits/{}".format(
                org_name, repo_name, rcommit.hexsha
            )

    else:
        origin = "git origin unknown"
    # deduplication if similar commits on different branches
    if rcommit.hexsha in potential_vulnerabilities:
        potential_vulnerabilities[rcommit.hexsha]["branches"].append(branch)
    else:
        potential_vulnerabilities[rcommit.hexsha] = {}
        potential_vulnerabilities[rcommit.hexsha]["message"] = rcommit.message
        try:
            lang = langdetect(rcommit.message)
        except:
            lang = "unknown"
        potential_vulnerabilities[rcommit.hexsha]["language"] = lang
        potential_vulnerabilities[rcommit.hexsha]["commit-id"] = rcommit.hexsha
        potential_vulnerabilities[rcommit.hexsha]["summary"] = rcommit.summary
        potential_vulnerabilities[rcommit.hexsha]["stats"] = rcommit.stats.total
        potential_vulnerabilities[rcommit.hexsha]["author"] = rcommit.author.name
        potential_vulnerabilities[rcommit.hexsha]["author-email"] = rcommit.author.email
        potential_vulnerabilities[rcommit.hexsha][
            "authored_date"
        ] = rcommit.authored_date
        potential_vulnerabilities[rcommit.hexsha][
            "committed_date"
        ] = rcommit.committed_date
        potential_vulnerabilities[rcommit.hexsha]["branches"] = []
        potential_vulnerabilities[rcommit.hexsha]["branches"].append(branch)
        potential_vulnerabilities[rcommit.hexsha]["pattern-selected"] = pattern.pattern
        potential_vulnerabilities[rcommit.hexsha]["pattern-matches"] = vuln_match
        potential_vulnerabilities[rcommit.hexsha]["origin"] = origin
        if origin_github_api:
            potential_vulnerabilities[commit.hexsha][
                "origin-github-api"
            ] = origin_github_api
        potential_vulnerabilities[rcommit.hexsha]["tags"] = []
        if tags_matching:
            if repo.commit(rcommit).hexsha in tagmap:
                potential_vulnerabilities[rcommit.hexsha]["tags"] = tagmap[
                    repo.commit(rcommit).hexsha
                ]
        if cve:
            potential_vulnerabilities[rcommit.hexsha]["cve"] = cve
            potential_vulnerabilities[rcommit.hexsha]["state"] = "cve-assigned"
        else:
            potential_vulnerabilities[rcommit.hexsha]["state"] = commit_state

    return rcommit.hexsha, potential_vulnerabilities, cve_found


def summary_event(
    commit,
    pattern,
    element,
    vuln_match=None,
    commit_state="under-review"
):
    potential_vulnerabilities = {}

    cve, cve_found = extract_cve(commit["message"])

    potential_vulnerabilities[commit["sha"]] = {}
    potential_vulnerabilities[commit["sha"]]["repo_name"] = element["repo"]["name"]
    potential_vulnerabilities[commit["sha"]]["message"] = commit["message"]
    potential_vulnerabilities[commit["sha"]]["language"] = langdetect(commit["message"])
    potential_vulnerabilities[commit["sha"]]["commit-id"] = commit["sha"]
    potential_vulnerabilities[commit["sha"]]["author"] = commit["author"]["name"]
    potential_vulnerabilities[commit["sha"]]["author-email"] = commit["author"]["email"]
    potential_vulnerabilities[commit["sha"]]["authored_date"] = element["created_at"]
    potential_vulnerabilities[commit["sha"]]["branches"] = element["payload"]["ref"]
    potential_vulnerabilities[commit["sha"]]["pattern-selected"] = pattern.pattern
    potential_vulnerabilities[commit["sha"]]["pattern-matches"] = vuln_match
    potential_vulnerabilities[commit["sha"]]["origin-github-api"] = commit["url"]
    if cve:
        potential_vulnerabilities[commit["sha"]]["cve"] = cve
        potential_vulnerabilities[commit["sha"]]["state"] = "cve-assigned"
    else:
        potential_vulnerabilities[commit["sha"]]["state"] = commit_state

    return potential_vulnerabilities, cve_found


def extract_cve(commit):
    cve_found = set()
    cve_find = re.compile(r"CVE-[1-2]\d{1,4}-\d{1,7}", re.IGNORECASE)
    m = cve_find.findall(commit)
    if m:
        for v in m:
            cve_found.add(v)
        return m, cve_found
    else:
        return None, set()